[all-commits] [llvm/llvm-project] 59162e: [AMDGPU] Skip buffer_wbl2 before atomic fence acquire

[Stanislav Mekhanoshin via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 59162e38590fbe194e2f5dc11bcfc02bffeb75fc
      https://github.com/llvm/llvm-project/commit/59162e38590fbe194e2f5dc11bcfc02bffeb75fc
  Author: Stanislav Mekhanoshin <Stanislav.Mekhanoshin at amd.com>
  Date:   2023-03-08 (Wed, 08 Mar 2023)

  Changed paths:
    M llvm/lib/Target/AMDGPU/SIMemoryLegalizer.cpp
    M llvm/test/CodeGen/AMDGPU/memory-legalizer-fence.ll

  Log Message:
  -----------
  [AMDGPU] Skip buffer_wbl2 before atomic fence acquire

Memory models for gfx90a and gfx940 do not require buffer_wbl2
before the fence for acquire ordering, but we do insert the full
release.

Fixes: SWDEV-386785

Differential Revision: https://reviews.llvm.org/D145524